Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A system comprising: at least one projector configured to project a plurality of projected patterns onto a scene, the projected patterns including a first projected pattern that comprises a plurality of first projected features, wherein the first projected pattern comprises a dense pattern; a camera configured to capture a plurality of images comprising a first detected pattern corresponding to a reflection of the first projected pattern; and one or more processors configured to: compute a first depth map of the scene based on the first projected pattern, the first detected pattern, and relative positions of the camera and the at least one projector; project, using the at least one projector, a second projected pattern comprising a plurality of second projected features onto a portion of the scene, wherein the second projected pattern comprises a sparse pattern as compared to the dense pattern, and capture, using the camera, a second detected pattern corresponding to a reflection of the second projected pattern; compute a second depth map of the portion of the scene based on the second projected pattern, the second detected pattern, and the relative positions of the camera and the at least one projector; and update the first depth map of the scene based on the second depth map of the portion of the scene.
This invention relates to a depth sensing system that uses structured light projection and imaging to create detailed 3D maps of a scene. The system addresses the challenge of accurately capturing depth information in dynamic or complex environments where traditional methods may fail due to occlusion, low contrast, or limited resolution. The system includes at least one projector that emits multiple patterns onto a scene, starting with a dense pattern containing numerous features. A camera captures reflections of these patterns, and processors analyze the detected patterns to compute an initial depth map based on the relative positions of the camera and projector. To refine accuracy, the system then projects a sparser pattern onto a specific portion of the scene, capturing and processing its reflection to generate a second depth map. This refined map is used to update and enhance the initial depth map, improving overall precision. The approach combines high-resolution initial mapping with targeted refinement, ensuring detailed and accurate depth information across the entire scene. The system is particularly useful in applications requiring real-time 3D reconstruction, such as robotics, augmented reality, and autonomous navigation.
2. The system of claim 1 , wherein the second projected pattern is sparse as compared to the first projected pattern when a distance between adjacent second projected features of the second projected pattern is greater than a distance between adjacent first projected features of the first projected pattern.
This invention relates to a structured light projection system used for three-dimensional (3D) imaging or depth sensing, addressing the challenge of balancing accuracy and computational efficiency in depth mapping. The system projects two distinct patterns onto a target surface to capture depth information. The first projected pattern is a dense pattern with closely spaced features, providing high-resolution depth data but requiring significant computational resources for processing. The second projected pattern is a sparse pattern, where the distance between adjacent features is greater than in the first pattern, reducing computational load while still enabling depth estimation. The system dynamically adjusts the pattern density based on the application requirements, optimizing between accuracy and processing efficiency. The sparse pattern may be used for coarse depth estimation or initial alignment, while the dense pattern provides fine details. This dual-pattern approach improves performance in real-time applications such as robotics, augmented reality, or industrial inspection, where both speed and precision are critical. The system may include a projector, a camera, and processing circuitry to analyze the reflected patterns and reconstruct the 3D structure of the target surface. The invention enhances existing structured light systems by introducing adaptability in pattern density to meet varying performance demands.
3. The system of claim 1 , wherein a distance between adjacent second projected features of the second projected pattern is based on a predetermined minimum object size, and the predetermined minimum object size is proportional to a size of one or more expected objects that are expected to appear in the portion of the scene.
This invention relates to a system for detecting objects in a scene using projected patterns. The system addresses the challenge of accurately identifying objects of varying sizes within a monitored area by dynamically adjusting the spacing of projected features based on the expected minimum object size. The system projects a first pattern onto a scene, captures an image of the scene with the first pattern, and processes the image to detect objects. A second pattern is then projected onto the scene, where the distance between adjacent features of the second pattern is determined by a predetermined minimum object size. This minimum object size is proportional to the size of one or more expected objects anticipated to appear in the monitored portion of the scene. By tailoring the feature spacing to the expected object sizes, the system improves detection accuracy and reduces false positives. The system may also include a light source for projecting the patterns and an imaging device for capturing the scene. The second pattern is projected onto the same portion of the scene as the first pattern, ensuring consistent coverage for object detection. This approach enhances the system's ability to reliably detect objects of varying sizes in dynamic environments.
4. The system of claim 3 , wherein the distance between adjacent second projected features is further based on a predetermined minimum distance by which the expected objects are expected to move in the portion of the scene.
This invention relates to a system for projecting features onto a scene to detect and track objects. The system addresses the challenge of accurately identifying and monitoring moving objects in a dynamic environment by dynamically adjusting the spacing of projected features based on the expected motion of objects. The system includes a projector that generates a pattern of first projected features, which are used to establish a reference grid or baseline for the scene. A sensor captures images of the scene, and a processor analyzes these images to detect objects by identifying deviations from the reference grid caused by the presence of objects. The system then projects a second set of features, which are more densely spaced in areas where objects are detected, improving tracking accuracy. The spacing between these second projected features is further adjusted based on a predetermined minimum distance corresponding to the expected movement of objects in the scene. This ensures that the projected features remain sufficiently close to track objects even as they move, while avoiding excessive overlap or redundancy. The system dynamically updates the feature spacing in real-time to maintain optimal tracking performance as objects move through the scene.
5. The system of claim 1 , wherein the second projected pattern is smaller than the first projected pattern, and the number of second projected features of the second projected pattern is less than the number of first projected features of the first projected pattern.
6. The system of claim 1 , wherein the projector is synchronized with the camera and configured to project the first and second projected patterns at different times, and the camera is configured to capture the first and second detected patterns at different times corresponding to the different times at which the first and second projected patterns are projected.
7. The system of claim 1 , wherein the projector is configured to project the first projected pattern onto one or more portions of the scene that are different from the portion onto which the second projected pattern is projected.
8. The system of claim 1 , wherein the projector is configured to project the second projected pattern onto different portions of the scene at different times.
9. The system of claim 1 , wherein the portion of the scene is selected so that the portion of the scene includes only objects that (a) have not moved or (b) have moved at less than a predetermined threshold rate of motion for at least a predetermined period of time.
This invention relates to a system for analyzing scenes, particularly for identifying and selecting portions of a scene where objects exhibit minimal or no motion. The system addresses the challenge of distinguishing between static and dynamic elements in a scene, which is useful in applications such as surveillance, autonomous navigation, or environmental monitoring where tracking or ignoring motion is critical. The system processes a scene to detect objects and assess their motion characteristics. It selects a portion of the scene where objects either (a) have not moved at all or (b) have moved at a rate below a predefined threshold for at least a specified duration. This selection ensures that only objects meeting these criteria are included in the analyzed portion, effectively filtering out transient or rapidly moving elements. The system may use motion detection algorithms, such as optical flow or frame differencing, to determine object movement over time. The predetermined threshold and time period can be adjusted based on application requirements, allowing flexibility in defining what constitutes "minimal" motion. By focusing on static or slowly moving objects, the system enables more accurate analysis of stable elements in the scene, such as stationary obstacles, background structures, or objects of interest that are not in motion. This capability is particularly valuable in scenarios where distinguishing between moving and non-moving objects is essential for decision-making or further processing.
10. The system of claim 1 , wherein the portion of the scene is selected based on: one or more random numbers, a particular application, or the detected pattern.
A system for selecting a portion of a scene in a visual processing application addresses the challenge of efficiently analyzing or processing specific regions of an image or video frame. The system dynamically determines which portion of the scene to focus on based on one or more selection criteria. These criteria include random number generation, application-specific requirements, or detected patterns within the scene. Random number selection allows for unbiased sampling, useful in applications like statistical analysis or machine learning training. Application-specific selection tailors the portion to the needs of the software, such as focusing on a region of interest in medical imaging or surveillance. Pattern detection enables the system to identify and prioritize areas with significant features, such as edges, textures, or objects, improving efficiency in tasks like object recognition or anomaly detection. The system integrates these selection methods to optimize processing resources and enhance accuracy in various visual analysis tasks.
11. The system of claim 1 , wherein the portion of the scene is adjacent to an identified side of a previous portion in which an object was identified in the detected pattern in a previous frame, wherein the identified side is closer to the object than other sides of the previous portion.
12. The system of claim 11 , wherein the portion of the scene is adjacent to an identified side of a previous portion in which an object was identified in the detected pattern in a previous frame, wherein the identified side is in a direction of motion of the object.
This invention relates to computer vision systems for tracking objects in sequential video frames. The problem addressed is accurately identifying and tracking objects as they move across frames, particularly when objects partially leave or enter the field of view. The system analyzes a scene by dividing it into portions and detects patterns within those portions to identify objects. In subsequent frames, the system focuses on portions adjacent to previously identified object locations, specifically on the side of the previous portion that aligns with the object's direction of motion. This approach improves tracking accuracy by predicting where an object is likely to appear next based on its movement trajectory. The system dynamically adjusts its analysis regions to follow the object's path, reducing errors caused by partial occlusions or rapid movements. This method is particularly useful in applications like surveillance, autonomous navigation, and augmented reality, where reliable object tracking is critical. The invention enhances existing computer vision techniques by incorporating motion direction into the tracking process, ensuring more consistent object detection across frames.
13. The system of claim 1 , wherein a size and location of the portion of the scene are determined so that: the entire scene is covered by successive portions of the scene after a predetermined number of frames, or objects of at least a threshold minimum size are detected after the predetermined number of frames.
14. The system of claim 1 , wherein to project the first projected pattern, the projector is further configured to project different determined portions of the first projected pattern at different times during a time interval, wherein the plurality of images captured by the camera are captured at different times corresponding to the different times at which the portions of the first projected pattern are projected, and the first detected pattern is based on the plurality of images, each of the images (a) comprising a detected portion of the first detected pattern and (b) corresponding to reflections of a corresponding determined portion of the first projected pattern.
This invention relates to a structured light projection system used for 3D scanning or depth sensing, addressing challenges in capturing high-resolution depth information from dynamic or reflective surfaces. The system includes a projector and a camera synchronized to project and capture structured light patterns. The projector emits a first projected pattern divided into multiple portions, each projected at distinct times within a time interval. The camera captures multiple images at corresponding times, each image containing a detected portion of the first detected pattern derived from reflections of the projected pattern portions. By analyzing these images, the system reconstructs the full detected pattern, enabling accurate depth mapping despite potential motion or surface irregularities. The approach improves robustness in environments where traditional single-pattern projections may fail due to interference or incomplete data. The system may also incorporate additional patterns or processing steps to enhance accuracy or handle complex scenes. This method is particularly useful in applications like industrial inspection, augmented reality, or autonomous navigation where precise 3D reconstruction is critical.
15. The system of claim 14 , wherein the determined portions of the first projected pattern are determined by dividing the first projected pattern into a specified number of portions.
The invention relates to a system for processing projected patterns, particularly in applications such as structured light projection for 3D scanning or depth sensing. The system addresses the challenge of accurately analyzing and interpreting projected patterns, which can be distorted or partially occluded in real-world environments. The system includes a projector that emits a first projected pattern onto a target surface, and a sensor that captures an image of the pattern as it interacts with the surface. The system then processes the captured image to determine specific portions of the first projected pattern, which are used for further analysis, such as depth calculation or surface reconstruction. These portions are identified by dividing the first projected pattern into a predefined number of segments or regions. The division may be based on spatial, intensity, or other pattern characteristics to ensure consistent and reliable analysis. The system may also adjust the division dynamically based on environmental conditions or pattern distortions to improve accuracy. This approach enhances the robustness of pattern-based sensing systems in applications like industrial inspection, medical imaging, or augmented reality.
16. The system of claim 15 , wherein the at least one projector comprises a plurality of light emitters, and the processors are further configured to: determine a number of emitters per portion by dividing the number of emitters of the projector by the specified number of determined portions of the first projected pattern, wherein the number of emitters in each determined portion of the first projected pattern is based on the determined number of emitters per portion.
A projection system is designed to enhance display flexibility by dynamically adjusting light emitters to create multiple distinct projected patterns. The system addresses the challenge of rigid, single-pattern projection by enabling a single projector to generate multiple independent patterns, improving adaptability for various applications such as augmented reality, interactive displays, or multi-zone lighting. The projector includes multiple light emitters, and a processing unit divides these emitters into portions corresponding to the desired number of projected patterns. The system calculates the number of emitters allocated to each pattern portion by dividing the total emitters by the specified number of patterns. This ensures balanced distribution of emitters across patterns, optimizing light output and resolution for each. The processing unit dynamically adjusts emitter allocation based on real-time requirements, allowing seamless switching between different projection configurations without hardware modifications. This approach enhances versatility, reduces cost, and simplifies system design by leveraging a single projector for multiple patterns.
17. The system of claim 14 , wherein a set of one or more adjacent emitters is assigned to each determined portion of the first projected pattern, and the adjacent emitters comprise physically-adjacent emitters or emitters that project adjacent first projected features of the first projected pattern.
This invention relates to a structured light projection system used for 3D scanning or depth sensing, addressing challenges in accurately projecting and detecting structured light patterns for precise depth measurement. The system projects a first pattern onto a target surface, where the pattern consists of multiple features (e.g., dots, lines, or grids) that are distorted by the surface's geometry. A sensor captures the distorted pattern, and the system analyzes the distortions to reconstruct the 3D shape of the surface. The system includes a light source with multiple emitters, each capable of projecting a portion of the first pattern. The emitters are arranged such that adjacent emitters either share a physical proximity or project adjacent features in the pattern. This ensures seamless pattern coverage without gaps or overlaps, improving measurement accuracy. The system dynamically assigns emitters to specific portions of the pattern based on the target surface's characteristics, optimizing projection efficiency and reducing errors caused by misalignment or environmental interference. The emitters may be modulated or controlled independently to enhance pattern clarity and contrast, further improving depth resolution. The system may also incorporate calibration techniques to compensate for emitter misalignment or variations in light intensity. This approach enables high-precision 3D scanning in applications such as industrial inspection, medical imaging, or autonomous navigation.
18. The system of claim 15 , wherein the specified number of portions of the first projected pattern is determined based on a rate of motion of one or more objects detected in the images captured by the camera.
19. One or more computer-readable non-transitory storage media embodying software that is operable when executed to: compute a first depth map of a scene based on a first projected pattern, a first detected pattern, and relative positions of a camera and at least one projector, wherein the at least one projector is configured to project a plurality of projected patterns onto the scene, the projected patterns include the first projected pattern, and the first projected pattern comprises a plurality of first projected features, wherein the first projected pattern comprises a dense pattern, and wherein the camera is configured to capture a plurality of images comprising the first detected pattern, and the first detected pattern corresponds to a reflection of the first projected pattern; project, using the at least one projector, a second projected pattern comprising a plurality of second projected features onto a portion of the scene, wherein the second projected pattern comprises a sparse pattern as compared to the dense pattern; capture, using the camera, a second detected pattern corresponding to a reflection of the second projected pattern; compute a second depth map of the portion of the scene based on the second projected pattern, the second detected pattern, and the relative positions of the camera and the at least one projector; and update the first depth map of the scene based on the second depth map of the portion of the scene.
This invention relates to a system for improving depth mapping accuracy in 3D scene reconstruction using structured light projection. The problem addressed is the trade-off between dense pattern projection, which provides high-resolution depth data but is computationally intensive, and sparse pattern projection, which is faster but less accurate. The solution involves a hybrid approach that combines both techniques to enhance depth map precision while maintaining efficiency. The system uses at least one projector to project multiple patterns onto a scene, including a dense pattern with numerous features for initial high-resolution depth mapping. A camera captures the reflected patterns, and a depth map is computed based on the projected and detected patterns, along with the relative positions of the camera and projector. To refine the depth map, a sparse pattern with fewer features is projected onto a specific portion of the scene. The camera captures this reflection, and a second depth map is computed for that portion. The initial depth map is then updated using the refined data from the sparse pattern projection, improving accuracy in the targeted area. This method allows for efficient, high-precision depth mapping by leveraging the strengths of both dense and sparse patterns.
20. A method comprising: by a computing device, computing a first depth map of a scene based on a first projected pattern, a first detected pattern, and relative positions of a camera and at least one projector, wherein the at least one projector is configured to project a plurality of projected patterns onto the scene, the projected patterns include the first projected pattern, and the first projected pattern comprises a plurality of first projected features, wherein the first projected pattern comprises a dense pattern, and wherein the camera is configured to capture a plurality of images comprising the first detected pattern, and the first detected pattern corresponds to a reflection of the first projected pattern; by the computing device, projecting, using the at least one projector, a second projected pattern comprising a plurality of second projected features onto a portion of the scene, wherein the second projected pattern comprises a sparse pattern as compared to the dense pattern; by the computing device, capturing, using the camera, a second detected pattern corresponding to a reflection of the second projected pattern; by the computing device, computing a second depth map of the portion of the scene based on the second projected pattern, the second detected pattern, and the relative positions of the camera and the at least one projector; and by the computing device, updating the first depth map of the scene based on the second depth map of the portion of the scene.
Unknown
January 19, 2021
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.